• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.07a
• /
• pp.9-10
• /
• 2019

본 논문은 구동장치의 진동에서 특징 데이터를 추출하고 인공신경망에 학습을 시킨 후, 구동 장치의 결함을 분류하는 시스템을 구현하였다. 딥러닝 기술을 이용함으로써 특정 장치에 종속되지 않고 학습할 데이터의 특징에 따라 쉽게 변경 가능하다. 또한, 실제 적용될 현장에서 발생할 수 있는 예측외의 진동 환경에 유연하게 대처하기 위해 딥러닝 모델 중 CNN을 적용한 시스템을 설계하였으며, 본 연구팀의 이전 연구에서 제안된 DNN 기반의 진단시스템을 학습데이터의 환경과 다른 처리배제가 필요한 진동 환경에서 비교 실험하여 제안된 시스템이 새로운 환경적응 성능향상에 대하여 우수한 결과를 얻었음을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1232-1235
• /
• 2022

최근 자동차, 철도차량 등 사용자가 있는 기계 시스템에서의 고장 발생 시 사용자의 안전과 관련된 사고로 이어질 수 있어 부품에 대한 모니터링 및 고장 여부 판단은 매우 중요하다. 이러한 부품 중에서 베어링은 회전체와 회전하지 않는 물체 사이에서 회전이 원활하게 이루어질 수 있도록 하는 부품인데, 베어링에 결함이 발생하게 될 경우, 기계 시스템이 정지하거나, 마찰 열에 의해 화재 등의 치명적인 위험이 발생한다. 본 논문에서는 Resnet과 오토인코더를 활용하여 진동 신호 기반의 베어링의 고장을 감지하고 분류할 수 있는 모델을 제안한다. 제안 방법은 raw data를 이미지로 변환하여 입력으로 사용하는데, 이러한 접근을 통해 수집된 데이터의 손실을 최소화하고 데이터가 가지는 정보를 최대한 분석에 활용할 수 있다. 제안 모델의 검증을 위하여 공개된 데이터셋으로 학습/검증 하였고, 제안 방법이 기존 방법과 비교하여 더 높은 F1 Score와 정확도를 보임을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1258-1260
• /
• 2022

재료공학에서 머신러닝을 이용해 목적 성능에 부합하는 물질의 조성을 탐색하는 연구가 있다. 물질의 성능은밀도 범함수 계산을 통해 시뮬레이션 할 수 있지만, 계산량이 많은 문제가 있다. 본 연구를 통해 우리는 고효율 페로브스카이트 태양광전지를 만들기 위한 페로브스카이트 조성을 추천하는 심층신경망과 베이지안 최적화 모델을 제안했다. 본 연구에서 높은 전력효율이 예상되는 페로브스카이트 조성을 심층신경망과 베이지안 최적화 방법을 통해 추천하는 모델을 구현하였다. 심층신경망 모델은 주어진 조성과 실험조건에서 예상되는 전력효율을 예측해 베이지안 최적화를 통한 탐색과정에서 소요되는 실험비용을 줄인다. 베이지안 최적화 모델은 실험공간을 입력으로 받아 고효율이 예상되는 실험조건을 출력하는데, 미리 설정한 실험공간만을 탐색하기 때문에 실험적으로 가능한 출력값만을 제시 할 수 있다. 본 연구는 심층신경망과 베이지안 최적화 방법을 조합해 주어진 실험공간을 탐색하는 시간과 비용을 최소화하는 방법을 제시한다

• Journal of Dental Rehabilitation and Applied Science
• /
• v.32 no.1
• /
• pp.38-46
• /
• 2016

Purpose: study for discriminating method of origin side vibration from non-symptomatic clicking group. Materials and Methods: 60 joints vibrations of 30 subjects in non-symptomatic clicking group was recorded via subject's awareness, examiner's palpation and JVA analysis. Origin side vibration was discriminated with consideration for frequency spectrum, time delay and phase shift of waveforms, analysis of numeric values. Results: There were all unilateral vibrations with JVA analysis and number of origin vibrations were 42. 11 pairs of vibrations showed time delay and phase shift and transferred side vibrations showed smaller values of total integral and bigger values of > 300 / < 300 ratio than origin side vibrations except one pair of vibrations. Also as the ipsi-lateral joint vibrations with smaller values of total integral showed bigger values of > 300 / < 300 ratio than the contra-lateral joint vibrations and there all ipsi-lateral vibrations were showed small values of total integral below 10 and hard to detect time delay and phase shift. So the features were used in discrimination of origin side vibrations. Conclusion: There should be all-around considerations for discrimination of origin side vibrations that is frequency spectrum, phase shift and time delay and analysis of numeric values.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.7
• /
• pp.23-35
• /
• 2013

This paper proposes an effective technique that is used to automatically extract feature vectors from vibration signals for fault classification systems. Conventional mel-frequency cepstral coefficients (MFCCs) are sensitive to noise of vibration signals, degrading classification accuracy. To solve this problem, this paper proposes spectral envelope cepstral coefficients (SECC) analysis, where a 4-step filter bank based on spectral envelopes of vibration signals is used: (1) a linear predictive coding (LPC) algorithm is used to specify spectral envelopes of all faulty vibration signals, (2) all envelopes are averaged to get general spectral shape, (3) a gradient descent method is used to find extremes of the average envelope and its frequencies, (4) a non-overlapped filter is used to have centers calculated from distances between valley frequencies of the envelope. This 4-step filter bank is then used in cepstral coefficients computation to extract feature vectors. Finally, a multi-layer support vector machine (MLSVM) with various sigma values uses these special parameters to identify faulty types of induction motors. Experimental results indicate that the proposed extraction method outperforms other feature extraction algorithms, yielding more than about 99.65% of classification accuracy.

• The Journal of Society for e-Business Studies
• /
• v.23 no.2
• /
• pp.111-121
• /
• 2018

Many problems in rotating machinery such as aircraft engines, wind turbines and motors are caused by bearing defects. The abnormalities of the bearing can be detected by analyzing signal data such as vibration or noise, proper pre-processing through a few signal processing techniques is required to analyze their frequencies. In this paper, we introduce the condition monitoring method for diagnosing the failure of the rotating machines by analyzing the vibration signal of the bearing. From the collected signal data, the normal states are trained, and then normal or abnormal state data are classified based on the trained normal state. For preprocessing, a Hamming window is applied to eliminate leakage generated in this process, and the cepstrum analysis is performed to obtain the original signal of the signal data, called the formant. From the vibration data of the IMS bearing dataset, we have extracted 6 statistic indicators using the cepstral coefficients and showed that the application of the Mahalanobis distance classifier can monitor the bearing status and detect the failure in advance.

• Tuberculosis and Respiratory Diseases
• /
• v.71 no.1
• /
• pp.30-36
• /
• 2011

Background: Pneumonia is commonly seen in outpatient clinics. it is widely known as the most common cause of death from infectious disease. Pneumonia has been diagnosed by its typical symptoms, chest X-ray and blood tests. However, both chest X-rays and blood tests have limitations in diagnosis. Thus primary care clinicians usually have been constrained due to a lack of adequate diagnostic tools. Vibration response imaging (VRI) is a newly emerging diagnostic modality, and its procedure is non-invasive, radiation-free, and easy to handle. This study was designed to evaluate the diagnostic usefulness of the VRI test among pneumonia patients and to consider its correlation with other conventional tests such as Chest X-ray, laboratory tests and clinical symptoms. Methods: VRI was performed in 46 patients diagnosed with pneumonia in Konkuk University Medical Center. VRI was assessed in a private and quiet room twice: before and after the treatment. Sensors for VRI were placed on a patient's back at regular intervals; they detected pulmonary vibration energy produced when respiration occurred and presented as specific images. Any modifications either in chest X-ray, C-reactive protein (CRP), white blood cell count (WBC) or body temperature were compared with changes in VRI image during a given time course. Results: VRI, chest X-ray and CRP scores were significantly improved after treatment. Correlation between VRI and other tests was not clearly indicated among all patients. But relatively severe pneumonia patients showed correlations between VRI and chest X-ray, as well as between VRI and CRP. Conclusion: This study demonstrates that VRI can be safely applied to patients with pneumonia.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.12 no.6
• /
• pp.57-63
• /
• 2012

Effective use of stethoscope is very important for primary clinical diagnosis for the increasing cardiovascular and respiratory disease. This study developed the contact vibration sensor using piezopolymer film which minimizes the ambient noise, and signal processing algorithm was applied for providing better auscultation sounds compare to the existing electronic stethoscopes. Especially, low frequency heart sounds were acquired without distortion, and the quality of lung sounds were improved. Also, auscultating sounds could be transmitted using bluetooth, which made possible to be used for the u-healthcare environment. Results of this study, auscultation of heart and lung sounds, could be applied to the convergence industry of medical and information communication technology through remote diagnosis.

• Journal of the Korean Society for Railway
• /
• v.20 no.2
• /
• pp.223-233
• /
• 2017

The railway turnout system is one of the most important systems that set train routes. Turnout system integrity should be guaranteed for robust train operation. To diagnose the turnout system status, LVDT and accelerometers are installed on a turnout system in a high speed line. The LVDT and accelerometers produce signals containing physical meaning of the turnout systems. The LVDT produces the displacement of the rail gauge and vibration when point moving or a train passes on turnout systems and the accelerometer produces impact forces induced by wheel sets. We performed data extraction from the measured signals and parameterized the extracted signals into meaningful quantities. The parameters are used for classifying whether the turnout status is normal. We proposed two methods for the classification, one uses probabilistic distribution and the other artificial neuron networks. The probabilistic distribution is used for the parameter being classified by the quantities and the artificial neuron networks for the form classification. Finally, we show how to learn the normal status of a turnout system.

• Journal of Institute of Control, Robotics and Systems
• /
• v.18 no.10
• /
• pp.921-927
• /
• 2012

In this paper we deal with a hydraulic tilting actuator to make a diagnosis of load characteristic acting on the tilting actuator of the tilting train. Tilting actuator in the tilting train plays a role of making tilt of the train when the train runs a curve section to make the train run without deceleration. However in the process of tilt the tilting actuator is affected by the load acting on the actuator, which has a possibility to make bogie vibration. In order to figure out the effect of the load on the tilting actuator a hydraulic tilting devices that are capable of tilting the train is proposed. The proposed devices are installed in the front bogie and in the rear bogie to make tilting of the train. The devices are consist of sensors that measure the load capacity of the actuator and displacement of the hydraulic cylinder stroke, control blocks to make synchronization of the two actuators, user interface block to monitor the status of the actuators. The effectiveness of the proposed hydraulic tilting actuators is presented by the experimental evaluation using actual tilting train.